Re: [RFC/PATCH] AVR32: Oprofile support

2007-11-08 Thread Philippe Elie
On Thu, 08 Nov 2007 at 12:14 +, Haavard Skinnemoen wrote: > > Why not a loop here ? > > The system register is coded into the instruction as an immediate, so > it would have to be a loop with a switch in it. And with only three > counters, I thought it looked cleaner (and faster) this way.

Re: [RFC/PATCH] AVR32: Oprofile support

2007-11-08 Thread Haavard Skinnemoen
On Thu, 8 Nov 2007 10:28:06 +0100 Philippe Elie <[EMAIL PROTECTED]> wrote: > On Wed, 07 Nov 2007 at 15:05 +, Haavard Skinnemoen wrote: > > + /* PCCNT */ > > + if (ctr->enabled && (pccr & ctr->flag_mask)) { > > + sysreg_write(PCCNT, -ctr->count); > > +

Re: [RFC/PATCH] AVR32: Oprofile support

2007-11-08 Thread Philippe Elie
On Wed, 07 Nov 2007 at 15:05 +, Haavard Skinnemoen wrote: > This adds the necessary architecture code to run oprofile on AVR32 > using the performance counters documented by the AVR32 Architecture > Manual. > > Signed-off-by: Haavard Skinnemoen <[EMAIL PROTECTED]> > --- /dev/null > +++

Re: [RFC/PATCH] AVR32: Oprofile support

2007-11-08 Thread Philippe Elie
On Wed, 07 Nov 2007 at 15:05 +, Haavard Skinnemoen wrote: This adds the necessary architecture code to run oprofile on AVR32 using the performance counters documented by the AVR32 Architecture Manual. Signed-off-by: Haavard Skinnemoen [EMAIL PROTECTED] --- /dev/null +++

Re: [RFC/PATCH] AVR32: Oprofile support

2007-11-08 Thread Philippe Elie
On Thu, 08 Nov 2007 at 12:14 +, Haavard Skinnemoen wrote: Why not a loop here ? The system register is coded into the instruction as an immediate, so it would have to be a loop with a switch in it. And with only three counters, I thought it looked cleaner (and faster) this way.

Re: [RFC/PATCH] AVR32: Oprofile support

2007-11-08 Thread Haavard Skinnemoen
On Thu, 8 Nov 2007 10:28:06 +0100 Philippe Elie [EMAIL PROTECTED] wrote: On Wed, 07 Nov 2007 at 15:05 +, Haavard Skinnemoen wrote: + /* PCCNT */ + if (ctr-enabled (pccr ctr-flag_mask)) { + sysreg_write(PCCNT, -ctr-count); + oprofile_add_sample(regs, PCCNT);

Re: [RFC/PATCH] AVR32: Oprofile support

2007-11-07 Thread Haavard Skinnemoen
On Wed, 7 Nov 2007 16:55:43 +0100 Philippe Elie <[EMAIL PROTECTED]> wrote: > On Wed, 07 Nov 2007 at 15:05 +, Haavard Skinnemoen wrote: > > > This adds the necessary architecture code to run oprofile on AVR32 > > using the performance counters documented by the AVR32 Architecture > > Manual.

Re: [RFC/PATCH] AVR32: Oprofile support

2007-11-07 Thread Philippe Elie
On Wed, 07 Nov 2007 at 15:05 +, Haavard Skinnemoen wrote: > This adds the necessary architecture code to run oprofile on AVR32 > using the performance counters documented by the AVR32 Architecture > Manual. Did you post the user space change to [EMAIL PROTECTED] ? -- regards, Phe - To

[RFC/PATCH] AVR32: Oprofile support

2007-11-07 Thread Haavard Skinnemoen
This adds the necessary architecture code to run oprofile on AVR32 using the performance counters documented by the AVR32 Architecture Manual. Signed-off-by: Haavard Skinnemoen <[EMAIL PROTECTED]> --- This patch is meant to go on top of Mathieu Desnoyers' "Move Instrumentation Support menu to

Re: [RFC/PATCH] AVR32: Oprofile support

2007-11-07 Thread Philippe Elie
On Wed, 07 Nov 2007 at 15:05 +, Haavard Skinnemoen wrote: This adds the necessary architecture code to run oprofile on AVR32 using the performance counters documented by the AVR32 Architecture Manual. Did you post the user space change to [EMAIL PROTECTED] ? -- regards, Phe - To

[RFC/PATCH] AVR32: Oprofile support

2007-11-07 Thread Haavard Skinnemoen
This adds the necessary architecture code to run oprofile on AVR32 using the performance counters documented by the AVR32 Architecture Manual. Signed-off-by: Haavard Skinnemoen [EMAIL PROTECTED] --- This patch is meant to go on top of Mathieu Desnoyers' Move Instrumentation Support menu to

Re: [RFC/PATCH] AVR32: Oprofile support

2007-11-07 Thread Haavard Skinnemoen
On Wed, 7 Nov 2007 16:55:43 +0100 Philippe Elie [EMAIL PROTECTED] wrote: On Wed, 07 Nov 2007 at 15:05 +, Haavard Skinnemoen wrote: This adds the necessary architecture code to run oprofile on AVR32 using the performance counters documented by the AVR32 Architecture Manual. Did you